Maenpuen et al. Appendix S1-7: Sources and consequences of mismatch between leaf disc and whole-leaf leaf mass per area (LMA)

Appendix S1:

Site information. MAT: mean annual temperature, MAP: mean annual precipitation, CV: coefficient of variation

Data name	Site name	Vegetation	Elevation (m)	Location	MAT (℃)	MAP (mm)	Dry period (month)	Canopy height (m)	Leaf thickness	Leaf disc thickness	Leaf punch diameter (cm)	Number of species for species means	Number of individuals for individuals means	Number of species for CV	Number of individuals for CV
Yunnan	Bubeng	Tropical forest (TRF)	780	21°37’N, 101°35’E	21	1532.0	6	35-45	Measured	Measured	0.6	60	366	60	366
Yunnan	Ailao-shan	Subtropical evergreen wet forest (STF)	2500	24°32’N, 101°01’E	11.7	1931.0		25	Measured	Measured	0.6	47	282	47	282
Yunnan	Yuanjiang	Hot-dry savanna (HDS)	480	23°28’N, 102°10’E	24.7	732.8	6	6-Apr	Measured	Measured	0.6	34	198	33	197
Yakushima	Yakushima	Warm-temperate forest	14 - 1748	30°38’N, 130°62’E	8.1 - 19.4	4477.0			Measured	Not-measured	1.0	193	519	5	26

The site information and climate data in Yunnan, referred to Fei et al. (2018), and Song et al. (2017). MAT and MAP of Yakushima were obtained from the Yakushima meteorological station and Eguchi (2006).

Appendix S2:

Detailed sampling protocol for whole-leaf and leaf disc LMA.

Appendix S3:

Generalization for a relationship between ratios of disc-based and whole-leaf estimates of leaf mass per area (LMA), leaf tissue density (LD), and leaf thickness (LT). For the i tree individual or species, the relationship between ratios of disc-based and whole-leaf estimates of LMA, LD, and LT is:

\[ \frac{LDd_i}{LDw_i} = \frac{LMAd_i}{LMAw_i} \frac{LTw_i}{LTd_i} \]

where d indicates disc-based estimates, w indicates whole-leaf based estimates.

Because thickness is measured on leaf lamina whether one uses a leaf disc or a whole-leaf, the expected ratio between thickness for a leaf disc and a whole-leaf should be 1. The above relationship, therefore, can be rewritten using lognormally distributed multiplicative error (\(\epsilon_i\)) on the arithmetic scale:

\[ \frac{LDd_i}{LDw_i} = \frac{LMAd_i}{LMAw_i} exp(\epsilon_i) \;\;\;\;\epsilon_i \sim N(0, \sigma^2). \]

LMA requires two times measurements (mass and area) and LD requires three time measurements (mass, area and thickness), and thus variance in the ratio of whole-leaf LD and leaf disc LD should be greater than that of LMA.

As we expected LD, showed the slightly smaller R² value than LMA (Figure below) in the Yunnan dataset. We have independent measurement for thickness for leaf discs and whole-leaves in the Yunnan dataset. If we use the same leaf thickness values for leaf discs and whole-leaves (i.e., LTw_i/LTd_i =1), the scatter plots will be identical for LMA and LD. Consequently, we do not perform further analyses for LD, because differences between whole-leaf LD and leaf disc LD only depends on the ratio between whole-leaf LMA and leaf disc LMA and measurement errors of leaf thickness.

Figure: Relationships between species mean leaf mass per area (LMA) and leaf tissue density (LD) determined by using whole leaves and leaf discs for the Yunnan dataset that has both leaf thickness for leaf discs and whole leaves. Dashed lines indicate 1:1 lines. Blue solid lines indicate standardized major axis regressions. The 95% confidence intervals are presented as the shaded area. All the correlations are significant (P < 0.001).

Appendix S4:

Standardized regression coefficients modelig the effects of leaf tissue density, leaf area, leaf thickness, punch size, and their interactios (a) on the mean estimates of whole-leaf LMA, and (b) on the estimated variance of whole-leaf LMA. The leaf disc LMA was used for the baseline mean of the whole-leaf LMA. Thick and thin lines indicate 90% and 95% credible intervals, respectively. Circles show posterior means of coefficients. Circles filled with blue indicate significant effects and white indicate non-significance effects.

Appendix S5:

Relationships between species mean leaf mass per area (LMA) determined by using whole-leaves and leaf discs for species obtained with leaf punches of different diameters. Dashed lines indicate 1:1 lines. Blue solid lines indicate standardized major axis (SMA) regressions. The 95% confidence intervals are resented as the shaded area. The correlation is significant (P < 0.001).

Appendix S6:

Relationships between whole-leaf LMA : disc leaf LMA and total dry mass for the leaf disc. The relationships was heteroscedastic (unequal variance, p < 0.001; all the samples were analyzed together). Samples with small total dry mass tended to show greater variance.

Appendix S7:

Summary of a linear mixed model for log-transformed total dry mass of leaf discs for each tree individual. Species was used as a random intercept. The leaf punch size is described as an effect of ‘0.6 cm diameter’ compared with ‘1.0 cm diameter’.

	Estimate	SE	DF	t-value	P value
(Intercept)	-4.41	0.013	1030	-333	< 0.001
log(Leaf tissue density)	0.24	0.007	1030	33.2	< 0.001
log(Leaf area)	-0.061	0.009	1030	-6.94	< 0.001
log(Leaf thickness)	0.277	0.008	1030	34.9	< 0.001
Small leaf punch	-0.5	0.02	332	-25.6	< 0.001

LITERATURE CITED

Eguchi, T. 2006. Climate. Yakushima, 3–26. Asakura Shoten, Tokyo (in Japanese).

Fei, X., Q. Song, Y. Zhang, Y. Liu, L. Sha, G. Yu, L. Zhang, et al. 2018. Carbon exchanges and their responses to temperature and precipitation in forest ecosystems in Yunnan, Southwest China. Science of The Total Environment 616–617: 824–840.

Song, Q.-H., Y.-P. Zhang, L.-Q. Sha, X.-B. Deng, Y. Deng, C.-S. Wu, Z.-Y. Lu, et al. 2017. Canopy temperature variability in a tropical rainforest, subtropical evergreen forest, and savanna forest in Southwest China. iForest - Biogeosciences and Forestry 10: 611.